“Anyone with OpenAI API access can implement this approach to create their own AI-assisted moderation system.”
OpenAI says GPT-4 cuts policy changes from months to hours and does six months of moderation work in about a day. Meta paid more than 11,000 moderators at least $1,000 each over trauma from the job. OpenAI’s own post admits the model carries training biases and needs humans checking it. The pitch is mental health relief. The product is API calls billed per post.